Distribution network cross-station coordinated control method, device and electronic equipment

Abstract

This invention provides a method, device, and electronic equipment for cross-distribution area collaborative control of a distribution network, belonging to the field of smart distribution network technology. The method includes: extracting spatiotemporal features from distribution network data of all distribution areas to obtain the spatiotemporal features of multiple distribution areas; constructing a cross-distribution area coupling coefficient matrix based on the distribution network data of all distribution areas; inputting the spatiotemporal features, the cross-distribution area coupling coefficient matrix, distribution network data, and environmental prediction data into a deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model; and determining a target reconfiguration scheme for the distribution network based on the local decision schemes of multiple distribution areas and the initial reconfiguration scheme, so that multiple distribution areas can perform collaborative control based on the target reconfiguration scheme. This invention addresses the problem of poor operational stability and reliability of traditional distribution networks under complex operating conditions, such as the intermittency and volatility of distributed energy sources and uneven load distribution across distribution areas.

Description

Technical Field

[0001] This invention relates to the field of smart distribution network technology, specifically to a method for cross-distribution area collaborative control of a distribution network, a device for cross-distribution area collaborative control of a distribution network, an electronic device, a machine-readable storage medium, and a computer program product. Background Technology

[0002] As the penetration rate of distributed energy (such as solar and wind power) in the distribution network continues to increase, and as users' diversified electricity demands grow, the operating characteristics of the distribution network are becoming increasingly complex.

[0003] Traditional distribution networks are mainly based on centralized power sources, with a relatively simple network structure and control strategies focused on ensuring power supply reliability. However, when faced with the intermittency and volatility of distributed energy sources, as well as uneven load distribution across different transformer areas, traditional distribution networks exhibit poor operational stability and reliability under complex operating conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a method, device, and electronic equipment for cross-distribution area collaborative control of a distribution network, in order to solve the problem of poor operational stability and reliability of traditional distribution networks under complex operating conditions when facing intermittent and fluctuating distributed energy sources and uneven loads across distribution areas.

[0005] To achieve the above objectives, embodiments of the present invention provide a method for cross-distribution area coordinated control of a distribution network, comprising:

[0006] Acquire sensor data from multiple transformer substations in the power distribution network; the sensor data from these multiple substations includes at least all distribution network data and environmental prediction data for all substations.

[0007] Spatiotemporal features are extracted based on the distribution network data of all the aforementioned distribution areas to obtain the spatiotemporal features of multiple distribution areas in the distribution network.

[0008] A cross-regional coupling coefficient matrix is ​​constructed based on the distribution network data of all the transformer substations. The cross-regional coupling coefficient matrix includes the cross-regional coupling coefficients of all transformer substations, and the cross-regional coupling coefficients characterize the sensitivity of the local transformer substation to voltage fluctuations caused by load changes in other transformer substations.

[0009] The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0010] Based on the local decision-making schemes of multiple distribution substations and the initial reconfiguration scheme of the distribution network, a target reconfiguration scheme for the distribution network is determined so that multiple distribution substations can perform coordinated control based on the target reconfiguration scheme.

[0011] Optionally, the cross-regional coupling coefficient is calculated based on the voltage change when the first region experiences voltage fluctuation, the load change of the remaining regions, and the per-unit value of the branch impedance between the first region and the remaining regions.

[0012] The first distribution area is any one of the distribution areas in the distribution network; the remaining distribution areas are all the distribution areas excluding the first distribution area.

[0013] Optionally, the cross-regional coupling coefficient is calculated using the following formula:

[0014] ;

[0015] in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change in the j-th transformer area; The per-unit value of the branch impedance between the i-th and j-th transformer substations is represented; n represents the number of the remaining transformer substations.

[0016] Optionally, the distribution network data for all transformer substations includes the topology data and operation data of all transformer substations; the step of extracting spatiotemporal features based on the distribution network data of all transformer substations to obtain the spatiotemporal features of multiple transformer substations includes:

[0017] Spatial features are obtained by using a graph convolutional neural network to extract spatial features from the topological data of all transformer areas.

[0018] Time features were extracted from the distribution network operation data of all transformer substations using a deep learning model.

[0019] The spatial and temporal features are fused using an attention fusion method to obtain the spatiotemporal features of multiple distribution network areas.

[0020] Optionally, the local decision-making scheme for each transformer area is obtained through the following steps:

[0021] Each edge agent in the control area generates a random waiting time through a trusted execution environment;

[0022] The edge agent with the shortest random waiting time is determined as the target edge agent, and the decision scheme of the target edge agent is determined as the local decision scheme of the transformer area.

[0023] Optionally, the local decision-making scheme of each edge agent is obtained by using a fuzzy logic controller based on the power distribution network operation data of the transformer area and the initial reconfiguration scheme of the power distribution network.

[0024] Optionally, the sensor data of the multiple transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all transformer substations.

[0025] The process of inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model includes:

[0026] The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0027] Optionally, determining the target reconfiguration scheme for the distribution network based on the local decision-making scheme for multiple transformer substations and the initial reconfiguration scheme for the distribution network includes:

[0028] When the initial reconfiguration scheme of the distribution network conflicts with the local decision-making scheme of the first distribution area, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network of the first distribution area.

[0029] In the event of an emergency in the first distribution area, the first distribution area shall prioritize handling the emergency before executing the distribution network target reconfiguration scheme; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution area is any one of the plurality of distribution areas.

[0030] On the other hand, embodiments of the present invention also provide a cross-regional coordinated control device for a distribution network, comprising:

[0031] The data acquisition module is used to acquire sensor data from multiple transformer substations in the power distribution network; the sensor data from these multiple substations includes at least the power distribution network data and environmental prediction data for all substations.

[0032] The model training and optimization module is used to extract spatiotemporal features based on the distribution network data of all the transformer substations to obtain the spatiotemporal features of multiple transformer substations in the distribution network.

[0033] The model evaluation module is used to construct a cross-regional coupling coefficient matrix based on the distribution network data of all the transformer substations. The cross-regional coupling coefficient matrix includes the cross-regional coupling coefficients of all transformer substations, and the cross-regional coupling coefficients characterize the sensitivity of the local transformer substation to voltage fluctuations caused by load changes in other transformer substations.

[0034] The upper-level global optimization module is used to input the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data and the environmental prediction data into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0035] The conflict resolution strategy module is used to determine the target reconfiguration scheme of the distribution network based on the local decision schemes of multiple distribution areas and the initial reconfiguration scheme of the distribution network, so that multiple distribution areas of the distribution network can perform coordinated control based on the target reconfiguration scheme of the distribution network.

[0036] Optionally, the cross-regional coupling coefficient is calculated based on the voltage change when the first region experiences voltage fluctuation, the load change of the remaining regions, and the per-unit value of the branch impedance between the first region and the remaining regions.

[0037] The first distribution area is any one of the distribution areas in the distribution network; the remaining distribution areas are all the distribution areas excluding the first distribution area.

[0038] Optionally, the cross-regional coupling coefficient is calculated using the following formula:

[0039] ;

[0040] in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change in the j-th transformer area; The per-unit value of the branch impedance between the i-th and j-th transformer substations is represented; n represents the number of the remaining transformer substations.

[0041] Optionally, the distribution network data for all transformer substations includes the topology data and operation data of all transformer substations; the step of extracting spatiotemporal features based on the distribution network data of all transformer substations to obtain the spatiotemporal features of multiple transformer substations includes:

[0042] Spatial features are obtained by using a graph convolutional neural network to extract spatial features from the topological data of all transformer areas.

[0043] Time features were extracted from the distribution network operation data of all transformer substations using a deep learning model.

[0044] The spatial and temporal features are fused using an attention fusion method to obtain the spatiotemporal features of multiple distribution network areas.

[0045] Optionally, the local decision-making scheme for each transformer area is obtained through the following steps:

[0046] Each edge agent in the control area generates a random waiting time through a trusted execution environment;

[0047] The edge agent with the shortest random waiting time is determined as the target edge agent, and the decision scheme of the target edge agent is determined as the local decision scheme of the transformer area.

[0048] Optionally, the local decision-making scheme of each edge agent is obtained by using a fuzzy logic controller based on the power distribution network operation data of the transformer area and the initial reconfiguration scheme of the power distribution network.

[0049] Optionally, the sensor data of the multiple transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all transformer substations.

[0050] The process of inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model includes:

[0051] The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0052] Optionally, determining the target reconfiguration scheme for the distribution network based on the local decision-making scheme for multiple transformer substations and the initial reconfiguration scheme for the distribution network includes:

[0053] When the initial reconfiguration scheme of the distribution network conflicts with the local decision-making scheme of the first distribution area, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network of the first distribution area.

[0054] In the event of an emergency in the first distribution area, the first distribution area shall prioritize handling the emergency before executing the distribution network target reconfiguration scheme; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution area is any one of the plurality of distribution areas.

[0055] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-mentioned cross-regional coordinated control method for power distribution networks.

[0056] On the other hand, the present invention also provides a machine-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned cross-distribution area coordinated control method for power distribution networks.

[0057] On the other hand, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-mentioned cross-distribution area collaborative control method for power distribution networks.

[0058] Through the above technical solution, this embodiment of the invention inputs the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model. Then, based on the local decision schemes of multiple distribution areas and the initial reconfiguration scheme of the distribution network, a target reconfiguration scheme of the distribution network is determined, so that multiple distribution areas of the distribution network can perform coordinated control based on the target reconfiguration scheme. This embodiment of the invention achieves efficient energy dispatching and precise voltage coordinated control across distribution areas through multimodal data fusion, deep reinforcement learning, and distributed collaborative decision-making. This improves the operational stability and reliability of the distribution network under complex operating conditions, particularly when facing intermittent and fluctuating distributed energy resources and uneven load distribution across distribution areas.

[0059] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0060] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings:

[0061] Figure 1 This is one of the flowcharts of the cross-distribution area collaborative control method for power distribution networks provided by the present invention;

[0062] Figure 2 This is the second flowchart of the cross-distribution area collaborative control method for power distribution networks provided by the present invention;

[0063] Figure 3 This is one of the structural schematic diagrams of the cross-distribution area collaborative control device for power distribution networks provided by the present invention;

[0064] Figure 4 This is the second schematic diagram of the structure of the cross-distribution area collaborative control device for power distribution networks provided by the present invention;

[0065] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0066] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0067] Method Implementation Examples

[0068] Please refer to Figure 1 This invention provides a method for cross-regional coordinated control of a distribution network, comprising:

[0069] Step 100: Obtain sensor data from multiple transformer substations in the power distribution network.

[0070] Electronic devices can acquire sensor data from multiple transformer substations in a power distribution network via a communication network. The transformer substations in a power distribution network can be understood based on the following scenario: a feeder in the power distribution network connects multiple transformer substations, each supplying power independently. A residential area or village is typically supplied by one or more transformer substations, each independently undertaking voltage conversion, power distribution, and operational monitoring tasks for users within its area. The sensor data from these multiple transformer substations includes at least all transformer substation data and environmental prediction data. In one embodiment, to consider multimodal data as much as possible, the sensor data from multiple transformer substations includes all transformer substation data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data. These sensors are distributed at various key locations in the power distribution network, such as busbars, feeders, and distributed energy access points, and can collect real-time topology data (reflecting the electrical connections between substations), power distribution network operation data (e.g., voltage, current, power), environmental prediction data (e.g., future light intensity, temperature, wind speed), user behavior data (e.g., electricity consumption patterns, load curves, electric vehicle charging patterns), and distributed energy output data, among other multi-source heterogeneous information.

[0071] Additionally, please refer to Figure 2 After data collection, electronic devices can also perform data preprocessing. For example, electronic devices can clean the collected data, remove noise and outliers, and perform data standardization, such as mean averaging, to give different types of data a uniform dimension and range.

[0072] Step 200: Extract spatiotemporal features based on the distribution network data of all the transformer substations to obtain the spatiotemporal features of multiple transformer substations in the distribution network.

[0073] The distribution network data for all transformer substations includes both topology data and operational data. The electronic equipment first extracts spatial features from the topology data of all substations to obtain spatial features; then it extracts temporal features from the operational data of all substations to obtain temporal features; finally, it fuses the spatial and temporal features to obtain the spatiotemporal features of multiple transformer substations in the distribution network.

[0074] In one embodiment, step 200, extracting spatiotemporal features based on the distribution network data of all transformer substations to obtain spatiotemporal features of multiple transformer substations, includes: extracting spatial features from the topological structure data of all transformer substations using a graph convolutional neural network to obtain spatial features; extracting temporal features from the distribution network operation data of all transformer substations using a deep learning model to obtain temporal features; and fusing the spatial features and the temporal features using an attention fusion method to obtain spatiotemporal features of multiple transformer substations in the distribution network.

[0075] In the topology data of all distribution network areas, each area is constructed with nodes (transformers) and edges (line connections) to form a graph. G =( V , E Node characteristics include voltage and load, and edge weights are electrical distance or impedance. The matrix representation is as follows: Adjacency Matrix A The distance matrix represents the connection between a node and its neighbors. D Calculate Euclidean distance based on node geographic coordinates (latitude and longitude).

[0076] In processing power distribution network topology data, electronic devices employ Graph Convolutional Networks (GCNs) to extract spatial features. Through multi-layer graph convolution operations, GCNs can aggregate information about nodes and their neighbors, thereby extracting spatial features reflecting the electrical connections between different distribution areas. Then, the electronic devices can use various deep learning models (e.g., the Transformer model) to extract temporal features, arranging power distribution network operation data (such as voltage, current, and power) in chronological order to form time-series data. In other embodiments, the electronic devices can also arrange power distribution network operation data (such as voltage, current, and power), environmental prediction data, and user electricity consumption behavior data in chronological order to form time-series data. Through a multi-head attention mechanism, the Transformer model can capture the dependencies between different time steps in the time-series data, thereby extracting temporal features. Finally, the electronic devices can use an attention fusion method to fuse the spatial features extracted by the GCNs and the temporal features extracted by the Transformer model. Specifically, firstly, the spatial and temporal features are linearly transformed and mapped to the same dimension, then the attention weights between the spatial and temporal features are calculated, and the spatial and temporal features are weighted and fused according to the attention weights. Thus, this embodiment of the invention generates a comprehensive spatiotemporal feature matrix of distribution network data for all distribution network areas in this way, providing rich input information for subsequent deep reinforcement learning models.

[0077] Step 300: Construct a cross-regional coupling coefficient matrix based on the distribution network data of all the transformer substations.

[0078] Electronic equipment constructs an inter-regional coupling coefficient matrix based on the distribution network data of all transformer substations. The inter-regional coupling coefficient matrix includes the inter-regional coupling coefficients of all transformer substations. These inter-regional coupling coefficients characterize the sensitivity of the local transformer substation to voltage fluctuations caused by load changes in other transformer substations.

[0079] In one embodiment, the cross-regional coupling coefficient is calculated based on the voltage change when a voltage fluctuation occurs in the first distribution area, the load change in the remaining distribution areas, and the per-unit value of the branch impedance between the first distribution area and the remaining distribution areas. The first distribution area is any one of all distribution areas in the distribution network; the remaining distribution areas are all distribution areas excluding the first distribution area. The voltage change when a voltage fluctuation occurs in the first distribution area can be the voltage change detected in one detection cycle. The load change in the remaining distribution areas can also be the load change detected in the aforementioned detection cycle. Specifically, for example, the cross-regional coupling coefficient is calculated using the following formula:

[0080] ;

[0081] in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change of the j-th transformer area; the load change can be characterized by the power change. This represents the per-unit value of the branch impedance between the i-th and j-th transformer substations. The per-unit value of the branch impedance is the ratio of the actual impedance (nominal value) between the i-th and j-th transformer substations to the selected reference impedance value. n represents the number of the remaining transformer substations, which is the difference between the total number of transformer substations and 1.

[0082] This invention quantifies the voltage sensitivity between distribution substations by calculating the cross-substation coupling coefficient, reflecting the degree to which load changes in one substation affect the voltage of other substations. By inputting this coefficient as a reference factor into a deep reinforcement learning model, the model can gain a more comprehensive understanding of the electrical connection characteristics of the distribution network and the interaction relationships between different substations.

[0083] Step 400: Input the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0084] Electronic devices aim to maximize overall system performance (e.g., minimizing voltage deviation, maximizing distributed energy absorption, optimizing energy storage charging and discharging strategies, etc.). The spatiotemporal features obtained in step 200, the cross-distribution area coupling coefficient matrix obtained in step 300, and the distribution network data and environmental prediction data from step 100 are input into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model. The distribution network data may include the global state of the distribution network (e.g., power distribution of each feeder, state of charge of energy storage, etc.). The environmental prediction data may include meteorological forecast data for the distribution network location over a future period (e.g., light intensity, temperature, wind speed, etc.). The distribution network reconfiguration scheme includes at least one of the following: the switching state sequence of all distribution areas in the distribution network (switches need to be opened and closed to adjust the grid topology), the charging and discharging power command of the energy storage system, and the output adjustment coefficient of the distributed energy source. In one embodiment, the distribution network reconfiguration scheme includes the switching state sequence of all distribution areas in the distribution network (switches need to be opened and closed to adjust the grid topology), the charging and discharging power command of the energy storage system, and the output adjustment coefficient of the distributed energy source.

[0085] In other embodiments, the sensor data from the multiple transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all substations. The spatiotemporal features, the cross-substation coupling coefficient matrix, the distribution network data, and the environmental prediction data are respectively input into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model. This includes: inputting the spatiotemporal features, the cross-substation coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data into the deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model. In one embodiment, the distribution network reconfiguration scheme includes the switch state sequence of all transformer substations in the distribution network (switches need to be opened and closed to adjust the grid topology), the charging and discharging power commands of the energy storage system, and the output adjustment coefficients of the distributed energy sources. This is achieved by considering multimodal data input into the deep reinforcement learning model, including spatiotemporal features, the cross-substation coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data. Deep reinforcement learning models can provide a more comprehensive understanding of the electrical connection characteristics of the distribution network and the interactions between different transformer substations. This is beneficial for further improving the operational stability and reliability of the distribution network under complex operating conditions.

[0086] In one embodiment, this invention defines the state space (hereinafter referred to as "state") as the historical spatiotemporal characteristics of the distribution network, the historical cross-distribution area coupling coefficient matrix, historical distribution network data, and historical environmental prediction data; and defines the possible action space (hereinafter referred to as "action") as the historical distributed energy output data of the distribution network and historical user electricity consumption behavior data, etc. The reward function (hereinafter referred to as "reward") is defined as maximizing the overall system performance, which can balance multiple objectives. For example, minimizing voltage deviation, maximizing distributed energy absorption, and optimizing energy storage charging and discharging strategies are all objectives. The deep reinforcement learning model outputs the switching state sequence of distribution areas (switches need to be opened and closed to adjust the grid topology), the charging and discharging power commands of the energy storage system, and the output adjustment coefficients of the distributed energy.

[0087] Deep reinforcement learning combines deep neural networks with Q-learning, using deep neural networks to approximate the Q-value function. In one embodiment, the deep reinforcement learning model of this invention is obtained through the following steps:

[0088] Initialize the parameters of the experience replay pool, deep Q-network, and deep reinforcement learning model;

[0089] Repeat the following steps until the set conditions are met:

[0090] Acquire status (i.e., sample fault data), and select actions based on an ε-greedy strategy (historical distributed energy output data and historical user electricity consumption behavior data, etc.):

[0091] Perform the action, observe the reward calculated based on the reward function, and the new state;

[0092] Store experiences (states, actions, rewards, and new states) in the replay pool;

[0093] A set number of experiences are randomly selected from the replay pool as samples;

[0094] Calculate the target Q-value and the predicted Q-value for each sample;

[0095] Based on the target Q-value and predicted Q-value for each sample, the loss function is calculated using mean squared error; based on the loss function, the model parameters of the deep Q-network are updated through backpropagation and an optimizer (such as Adam);

[0096] The model parameters of the deep Q-network are copied to the deep reinforcement learning model every fixed number of steps (e.g., every 100 training steps).

[0097] Thus, this embodiment of the invention achieves a power distribution network reconfiguration scheme for all distribution substations based on a deep reinforcement learning model.

[0098] Step 500: Based on the local decision-making schemes of multiple distribution substations and the initial reconfiguration scheme of the distribution network, determine the target reconfiguration scheme of the distribution network so that multiple distribution substations can perform coordinated control based on the target reconfiguration scheme of the distribution network.

[0099] Electronic devices can collaboratively consider local decision-making schemes for multiple transformer substations and the initial reconfiguration scheme of the distribution network to determine the target reconfiguration scheme for the distribution network. The local decision-making scheme for each transformer substation can be determined by an edge agent (such as an intelligent control unit integrated in a smart meter or distributed energy inverter) within the substation. Based on locally collected distribution network operation data (such as local voltage deviation and load fluctuation) and the reconfiguration scheme obtained in the previous steps, a fuzzy logic controller quickly makes a local response decision, resulting in a local decision-making scheme. In this embodiment, when conflicts exist between the local decision-making schemes of multiple transformer substations and the initial reconfiguration scheme of the distribution network, a conflict resolution strategy is used to determine the target reconfiguration scheme for the distribution network. The electronic devices issue commands based on the target reconfiguration scheme, sending control commands generated by the target reconfiguration scheme (such as switch opening / closing commands, reactive power compensation equipment switching commands, and distributed energy output adjustment commands) to the corresponding execution devices.

[0100] In one embodiment, determining a target reconfiguration scheme for the distribution network based on local decision-making schemes for multiple distribution substations and the initial reconfiguration scheme for the distribution network includes: when the initial reconfiguration scheme for the distribution network conflicts with the local decision-making scheme for the first distribution substation, determining the initial reconfiguration scheme for the distribution network as the target reconfiguration scheme for the first distribution substation; wherein, in the event of an emergency in the first distribution substation, the first distribution substation prioritizes handling the emergency event before executing the target reconfiguration scheme for the distribution network; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution substation is any one of the multiple distribution substations.

[0101] In this embodiment of the invention, when a conflict arises between the initial reconfiguration scheme of the distribution network and the local decision-making scheme of the first transformer substation, a conflict resolution strategy is adopted. The global command of the initial reconfiguration scheme of the distribution network has the highest priority and is executed first. For example, in one implementation scenario, the initial reconfiguration scheme of the distribution network aims to increase the photovoltaic absorption rate (target: absorption rate > 95%), instructing the photovoltaic inverter in transformer substation C to operate at full power (1MW). However, the local decision-making scheme of transformer substation C is that the voltage at the end of transformer substation C is close to 80% of the upper limit, and transformer substation C needs to force the photovoltaic inverter to operate at reduced capacity. At this time, a conflict arises between the initial reconfiguration scheme of the distribution network and the local decision-making scheme of transformer substation C. Since the global command of the initial reconfiguration scheme of the distribution network has the highest priority, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network, and thus the strategy of full power output (1MW) of the photovoltaic inverter in transformer substation C is executed first. Therefore, in this embodiment of the invention, when a conflict arises between the initial reconfiguration scheme of the distribution network and the local decision-making schemes of each transformer substation, the conflict is handled according to the rule that the global command of the initial reconfiguration scheme of the distribution network has a higher priority than the local decision-making scheme.

[0102] It should be noted that if an emergency event occurs before the initial distribution network reconfiguration plan is implemented in transformer substation C, such as voltage exceeding limits (±5% threshold) and / or equipment overload (110% rated current), transformer substation C will prioritize handling the emergency event before implementing the full-power output (1MW) strategy for its photovoltaic inverters. This embodiment of the invention ensures the safe and stable operation of the power grid under various complex conditions by prioritizing the handling of emergency events in the local transformer substation before implementing the overall initial distribution network reconfiguration plan.

[0103] This invention incorporates the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain an initial distribution network reconfiguration scheme output by the deep reinforcement learning model. Then, based on the local decision-making schemes of multiple distribution areas and the initial distribution network reconfiguration scheme, a target distribution network reconfiguration scheme is determined, enabling multiple distribution areas to perform coordinated control based on the target reconfiguration scheme. This invention utilizes multimodal data fusion, deep reinforcement learning, and distributed collaborative decision-making techniques to achieve efficient cross-regional energy dispatch and precise voltage coordinated control. This improves the operational stability and reliability of the distribution network under complex operating conditions, particularly when facing intermittent and fluctuating distributed energy resources and uneven cross-regional loads.

[0104] In other aspects of the embodiments of the present invention, the local decision-making scheme of each transformer area is obtained through the following steps: controlling each edge agent of the transformer area to generate a random waiting time through a trusted execution environment; determining the edge agent with the shortest random waiting time as the target edge agent; and determining the decision-making scheme of the target edge agent as the local decision-making scheme of the transformer area.

[0105] In this embodiment of the invention, a consensus algorithm is used to achieve efficient coordination among multiple edge agents in each distribution area. Edge agents generate random waiting times through a trusted execution environment, and the edge agent with the shortest waiting time submits its local decision-making scheme first. Each edge agent's local decision-making scheme is obtained using a fuzzy logic controller based on the distribution network operation data of the distribution area and the initial reconfiguration scheme of the distribution network.

[0106] Specifically, each edge agent within the same distribution area (such as the intelligent control unit integrated in smart meters, distributed energy inverters, etc.) generates a random waiting time through a trusted execution environment. The edge agent with the shortest waiting time submits its local decision-making solution first, becoming the temporary "leader." Other edge agents, while waiting, verify the validity and rationality of their submitted local decision-making solutions. If no better proposal is received within the time limit, the local decision-making solution is accepted, avoiding conflicts caused by multiple edge agents submitting proposals simultaneously. This embodiment of the invention ensures that the local decision-making solutions of multiple edge agents can be executed efficiently and consistently, and that the system can still operate normally even when faced with failures or malicious behavior from some edge agents, achieving a fault tolerance rate of 33%.

[0107] In other aspects of this invention, the cross-distribution area coordinated control method for power distribution networks further includes: continuously monitoring the operating status of the power distribution network and feeding back the actual execution results and the latest operating data to electronic equipment. Based on the feedback data, the control effect is evaluated. If the expected goals are not achieved, such as voltage deviation in the power distribution network or insufficient distributed energy absorption rate, the process returns to steps 200 and 300 for data processing and feature extraction to optimize and adjust the model and decisions. Thus, this invention forms a closed-loop control, thereby continuously improving the effectiveness of cross-distribution area energy dispatch and voltage coordinated control.

[0108] Device Examples

[0109] Please refer to Figure 3 On the other hand, embodiments of the present invention also provide a cross-regional collaborative control device for a distribution network. This device includes a data acquisition module 301, a model training and optimization module 302, a model evaluation module 303, an upper-level global optimization module (or upper-level global optimizer) 304, and a conflict resolution strategy module 305.

[0110] The data acquisition module is used to acquire sensor data from multiple transformer substations within the distribution network. This sensor data includes at least all distribution network data and environmental prediction data from each substation. It possesses data acquisition capabilities, obtaining data from various sensors via a communication network. These sensors are distributed at key locations within the distribution network, such as busbars, feeders, and distributed energy access points, enabling real-time acquisition of multi-source heterogeneous information, including distribution network operation data (such as voltage, current, and power), environmental prediction data (light intensity, temperature, and wind speed), user behavior data (electricity consumption patterns, load curves, and electric vehicle charging patterns), and distributed energy output data.

[0111] The model training and optimization module is used to extract spatiotemporal features based on the distribution network data of all the transformer substations, obtaining the spatiotemporal features of multiple transformer substations. Specifically, the model training and optimization module fuses multi-source data from multiple transformer substations to generate a comprehensive and accurate spatiotemporal feature matrix for data feature extraction.

[0112] The model evaluation module is used to construct a cross-regional coupling coefficient matrix based on the distribution network data of all the transformer substations. The cross-regional coupling coefficient matrix includes the cross-regional coupling coefficients of all transformer substations, and the cross-regional coupling coefficients characterize the sensitivity of the local transformer substation to voltage fluctuations caused by load changes in other transformer substations.

[0113] The upper-level global optimization module is used to input the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into the deep reinforcement learning model, respectively, to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model. In other words, the upper-level global optimization module outputs the target reconfiguration scheme of the distribution network based on the deep reinforcement learning model.

[0114] The conflict resolution strategy module is used to determine the target reconfiguration scheme for the distribution network based on the local decision schemes of multiple distribution areas and the initial reconfiguration scheme of the distribution network, so that multiple distribution areas can perform coordinated control based on the target reconfiguration scheme. Specifically, when a conflict occurs in the decision-making process, it is handled according to rules such as the global command having higher priority than the local decision scheme, and the local decision prioritizes handling emergency situations such as voltage over-limit and / or equipment overload.

[0115] In some embodiments, the cross-regional coupling coefficient is calculated based on the voltage change when the first region experiences voltage fluctuations, the load change of the remaining regions, and the per-unit value of the branch impedance between the first region and the remaining regions.

[0116] The first distribution area is any one of the distribution areas in the distribution network; the remaining distribution areas are all the distribution areas excluding the first distribution area.

[0117] Optionally, the cross-regional coupling coefficient is calculated using the following formula:

[0118] ;

[0119] in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change in the j-th transformer area; The per-unit value of the branch impedance between the i-th and j-th transformer substations is represented; n represents the number of the remaining transformer substations.

[0120] In some embodiments, the distribution network data for all transformer substations includes the topology data and operation data of all transformer substations; the step of extracting spatiotemporal features based on the distribution network data for all transformer substations to obtain the spatiotemporal features of multiple transformer substations includes:

[0121] Spatial features are obtained by using a graph convolutional neural network to extract spatial features from the topological data of all transformer areas.

[0122] Time features were extracted from the distribution network operation data of all transformer substations using a deep learning model.

[0123] The spatial and temporal features are fused using an attention fusion method to obtain the spatiotemporal features of multiple distribution network areas.

[0124] In some embodiments, the local decision-making scheme for each transformer area is obtained through the following steps:

[0125] Each edge agent in the control area generates a random waiting time through a trusted execution environment;

[0126] The edge agent with the shortest random waiting time is determined as the target edge agent, and the decision scheme of the target edge agent is determined as the local decision scheme of the transformer area.

[0127] In some embodiments, the local decision-making scheme of each edge agent is obtained by using a fuzzy logic controller based on the power distribution network operation data of the transformer area and the initial reconfiguration scheme of the power distribution network.

[0128] In some embodiments, the sensor data of the plurality of transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all transformer substations.

[0129] The process of inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model includes:

[0130] The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

[0131] In some embodiments, determining the target reconfiguration scheme for the distribution network based on the local decision-making scheme for multiple transformer substations and the initial reconfiguration scheme for the distribution network includes:

[0132] When the initial reconfiguration scheme of the distribution network conflicts with the local decision-making scheme of the first distribution area, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network of the first distribution area.

[0133] In the event of an emergency in the first distribution area, the first distribution area shall prioritize handling the emergency before executing the distribution network target reconfiguration scheme; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution area is any one of the plurality of distribution areas.

[0134] The cross-regional collaborative control device for the power distribution network includes a processor and a memory. The aforementioned data acquisition module 301, model training and optimization module 302, model evaluation module 303, upper-level global optimization module 304, and conflict resolution strategy module 305 are all stored as program units in the memory. The processor executes the aforementioned program units stored in the memory to achieve the corresponding functions.

[0135] A processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured.

[0136] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0137] The cross-distribution area collaborative control device for distribution networks aims to solve the dynamic optimization and collaborative control problems of distribution networks with a high proportion of distributed energy access. Through multimodal data fusion, deep reinforcement learning, and distributed collaborative decision-making technologies, it achieves efficient energy dispatching and precise voltage collaborative control across distribution areas. Please refer to... Figure 4 In one embodiment, the cross-distribution area collaborative control device for the distribution network adopts a four-layer architecture, including a perception layer, a platform layer, an application layer, and an execution layer. The perception layer includes the aforementioned data acquisition module, data processing module, and communication interface. The platform layer includes the aforementioned model training and optimization module, model evaluation module, security and privacy protection module, and data twin module. The application layer includes the aforementioned upper-layer global optimization module, lower-layer edge agent module, consensus algorithm module, and conflict resolution strategy module. The upper-layer global optimization module and the lower-layer edge agent module constitute a two-layer optimization decision module. The execution layer refers to execution devices such as feeder switches, reactive power compensation devices, and load controllers. It receives the distribution network target reconfiguration method from the application layer and executes specific control operations.

[0138] The data processing module performs preliminary processing on the data collected by the data acquisition module, such as data cleaning and format conversion. The communication interface transmits the collected data to the platform layer via a high-speed communication network (such as 5G or fiber optic). The security and privacy protection module ensures the security of multi-source data transmission through encrypted communication protocols, preventing data theft or tampering. The data twin module uses advanced simulation technology to build a virtual model highly consistent with the physical power distribution network, mapping the physical power grid state in real time and simulating the operation of the power distribution network under various extreme scenarios (such as typhoons, rainstorms, and load surges), providing a virtual environment for the formulation and verification of optimization strategies. The lower-layer edge agent module is used for local decision-making, making rapid local responses based on real-time data of local voltage deviation and load fluctuations to obtain local decision solutions. The consensus algorithm module enables efficient coordination among multiple edge agents; edge agents generate random waiting times through a trusted execution environment, with the node with the shortest waiting time submitting its decision first.

[0139] Therefore, the cross-distribution area collaborative control device of the present invention solves the shortcomings of existing distribution network technology in terms of dynamic adaptability, multi-source data utilization and cross-distribution area collaborative control, realizes real-time perception and precise control of the distribution network operation status, improves the distributed energy absorption rate, ensures voltage quality, enhances the resilience and reliability of the distribution network under complex operating conditions, and improves the overall operating efficiency and economy.

[0140] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call logic instructions in the memory 530 to execute a cross-regional cooperative control method for the distribution network. This method includes: acquiring sensor data from multiple distribution network areas; the sensor data from the multiple distribution network areas includes at least distribution network data and environmental prediction data for all distribution network areas; extracting spatiotemporal features based on the distribution network data from all distribution network areas to obtain spatiotemporal features of the multiple distribution network areas; constructing a cross-regional coupling coefficient matrix based on the distribution network data from all distribution network areas; the cross-regional coupling coefficient matrix includes cross-regional coupling coefficients for all distribution network areas, and the cross-regional coupling coefficients characterize the sensitivity of the local distribution network to voltage fluctuations caused by load changes in other distribution network areas; inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain an initial reconfiguration scheme for the distribution network output by the deep reinforcement learning model; and determining a target reconfiguration scheme for the distribution network based on the local decision schemes of the multiple distribution network areas and the initial reconfiguration scheme, so that the multiple distribution network areas can perform cooperative control based on the target reconfiguration scheme.

[0141] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0142] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a machine-readable storage medium. When the computer program is executed by a processor, the computer can execute a method for cross-regional coordinated control of a distribution network. The method includes: acquiring sensor data from multiple distribution network areas; the sensor data from the multiple distribution network areas includes at least distribution network data and environmental prediction data from all distribution network areas; extracting spatiotemporal features based on the distribution network data from all distribution network areas to obtain spatiotemporal features of the multiple distribution network areas; constructing a cross-regional coupling coefficient matrix based on the distribution network data from all distribution network areas; the cross-regional coupling coefficient matrix includes cross-regional coupling coefficients from all distribution network areas, and the cross-regional coupling coefficients characterize the sensitivity of local distribution network voltage fluctuations caused by load changes in other distribution network areas; inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain an initial reconfiguration scheme for the distribution network output by the deep reinforcement learning model; and determining a target reconfiguration scheme for the distribution network based on the local decision schemes of the multiple distribution network areas and the initial reconfiguration scheme for the distribution network, so that the multiple distribution network areas can perform coordinated control based on the target reconfiguration scheme for the distribution network.

[0143] In another aspect, the present invention also provides a machine-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for cross-distribution network coordinated control. This method includes: acquiring sensor data from multiple distribution network areas; the sensor data from the multiple distribution network areas includes at least distribution network data and environmental prediction data for all distribution network areas; extracting spatiotemporal features based on the distribution network data from all distribution network areas to obtain spatiotemporal features of the multiple distribution network areas; constructing a cross-distribution network coupling coefficient matrix based on the distribution network data from all distribution network areas; the cross-distribution network coupling coefficient matrix includes cross-distribution network coupling coefficients for all distribution network areas, the cross-distribution network coupling coefficients characterizing the sensitivity of local distribution network voltage fluctuations caused by load changes in other distribution network areas; inputting the spatiotemporal features, the cross-distribution network coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain an initial reconfiguration scheme for the distribution network output by the deep reinforcement learning model; and determining a target reconfiguration scheme for the distribution network based on the local decision schemes of the multiple distribution network areas and the initial reconfiguration scheme, so that the multiple distribution network areas can perform coordinated control based on the target reconfiguration scheme.

[0144] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0145] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for cross-distribution area coordinated control of a distribution network, characterized in that, include: Acquire sensor data from multiple transformer substations in the power distribution network; The sensor data from the multiple transformer substations includes at least the power distribution network data and environmental prediction data for all substations. Spatiotemporal features are extracted based on the distribution network data of all the aforementioned distribution areas to obtain the spatiotemporal features of multiple distribution areas in the distribution network. Construct a cross-regional coupling coefficient matrix based on the distribution network data of all the aforementioned distribution areas; The cross-regional coupling coefficient matrix includes the cross-regional coupling coefficients of all regions, and the cross-regional coupling coefficients characterize the sensitivity of the local region to voltage fluctuations caused by load changes in other regions. The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model. Based on the local decision-making schemes of multiple distribution areas and the initial reconfiguration scheme of the distribution network, a target reconfiguration scheme for the distribution network is determined so that multiple distribution areas of the distribution network can perform coordinated control based on the target reconfiguration scheme. The cross-regional coupling coefficient is calculated based on the voltage change when the first region experiences voltage fluctuation, the load change of the remaining regions, and the per-unit value of the branch impedance between the first region and the remaining regions. The first distribution area is any one of the distribution areas in the distribution network; the remaining distribution areas are all the distribution areas excluding the first distribution area.

2. The cross-distribution area coordinated control method for power distribution networks according to claim 1, characterized in that, The cross-regional coupling coefficient is calculated using the following formula: ； in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change in the j-th transformer area; The per-unit value of the branch impedance between the i-th and j-th transformer substations is represented; n represents the number of the remaining transformer substations.

3. The cross-distribution area coordinated control method for power distribution networks according to claim 1, characterized in that, The distribution network data for all transformer substations includes the topology data and operation data of all transformer substations; the spatiotemporal feature extraction based on the distribution network data of all transformer substations yields the spatiotemporal features of multiple transformer substations, including: Spatial features are obtained by using a graph convolutional neural network to extract spatial features from the topological data of all transformer areas. Time features were extracted from the distribution network operation data of all transformer substations using a deep learning model. The spatial and temporal features are fused using an attention fusion method to obtain the spatiotemporal features of multiple distribution network areas.

4. The cross-distribution area coordinated control method for power distribution networks according to claim 3, characterized in that, The local decision-making plan for each transformer area is obtained through the following steps: Each edge agent in the control area generates a random waiting time through a trusted execution environment; The edge agent with the shortest random waiting time is determined as the target edge agent, and the decision scheme of the target edge agent is determined as the local decision scheme of the transformer area.

5. The cross-distribution area coordinated control method for power distribution networks according to claim 4, characterized in that, Each edge agent's local decision-making scheme is obtained by using a fuzzy logic controller based on the distribution network operation data of the transformer area and the initial reconfiguration scheme of the distribution network.

6. The cross-distribution area coordinated control method for distribution networks according to claim 1, characterized in that, The sensor data from the multiple transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all substations. The process of inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model includes: The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

7. The cross-distribution area coordinated control method for distribution networks according to claim 1, characterized in that, The local decision-making scheme based on multiple transformer substations and the initial reconfiguration scheme of the distribution network determine the target reconfiguration scheme of the distribution network, including: When the initial reconfiguration scheme of the distribution network conflicts with the local decision-making scheme of the first distribution area, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network of the first distribution area. In the event of an emergency in the first distribution area, the first distribution area shall prioritize handling the emergency before executing the distribution network target reconfiguration scheme; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution area is any one of the plurality of distribution areas.

8. A cross-distribution area collaborative control device for a power distribution network, characterized in that, include: The data acquisition module is used to acquire sensor data from multiple transformer substations in the power distribution network. The sensor data from the multiple transformer substations includes at least the power distribution network data and environmental prediction data for all substations. The model training and optimization module is used to extract spatiotemporal features based on the distribution network data of all the transformer substations to obtain the spatiotemporal features of multiple transformer substations in the distribution network. The model evaluation module is used to construct a cross-regional coupling coefficient matrix based on the distribution network data of all the aforementioned distribution areas; The cross-regional coupling coefficient matrix includes the cross-regional coupling coefficients of all regions, and the cross-regional coupling coefficients characterize the sensitivity of the local region to voltage fluctuations caused by load changes in other regions. The upper-level global optimization module is used to input the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data and the environmental prediction data into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model. The conflict resolution strategy module is used to determine the target reconfiguration scheme of the distribution network based on the local decision schemes of multiple distribution areas and the initial reconfiguration scheme of the distribution network, so that multiple distribution areas of the distribution network can perform coordinated control based on the target reconfiguration scheme of the distribution network. The cross-regional coupling coefficient is calculated based on the voltage change when the first region experiences voltage fluctuation, the load change of the remaining regions, and the per-unit value of the branch impedance between the first region and the remaining regions. The first distribution area is any one of the distribution areas in the distribution network; the remaining distribution areas are all the distribution areas excluding the first distribution area.

9. The cross-regional coordinated control device for power distribution networks according to claim 8, characterized in that, The cross-regional coupling coefficient is calculated using the following formula: ； in, C i This represents the cross-region coupling coefficient of the i-th transformer region; This represents the voltage change in the i-th transformer area; This represents the load change in the j-th transformer area; The per-unit value of the branch impedance between the i-th and j-th transformer substations is represented; n represents the number of the remaining transformer substations.

10. The cross-regional coordinated control device for power distribution networks according to claim 8, characterized in that, The distribution network data for all transformer substations includes the topology data and operation data of all transformer substations; the spatiotemporal feature extraction based on the distribution network data of all transformer substations yields the spatiotemporal features of multiple transformer substations, including: Spatial features are obtained by using a graph convolutional neural network to extract spatial features from the topological data of all transformer areas. Time features were extracted from the distribution network operation data of all transformer substations using a deep learning model. The spatial and temporal features are fused using an attention fusion method to obtain the spatiotemporal features of multiple distribution network areas.

11. The cross-regional coordinated control device for power distribution networks according to claim 10, characterized in that, The local decision-making plan for each transformer area is obtained through the following steps: Each edge agent in the control area generates a random waiting time through a trusted execution environment; The edge agent with the shortest random waiting time is determined as the target edge agent, and the decision scheme of the target edge agent is determined as the local decision scheme of the transformer area.

12. The cross-regional coordinated control device for power distribution networks according to claim 11, characterized in that, Each edge agent's local decision-making scheme is obtained by using a fuzzy logic controller based on the distribution network operation data of the transformer area and the initial reconfiguration scheme of the distribution network.

13. The cross-regional coordinated control device for power distribution networks according to claim 8, characterized in that, The sensor data from the multiple transformer substations includes distribution network data, environmental prediction data, distributed energy output data, and user electricity consumption behavior data for all substations. The process of inputting the spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, and the environmental prediction data into a deep reinforcement learning model to obtain the initial distribution network reconfiguration scheme output by the deep reinforcement learning model includes: The spatiotemporal features, the cross-regional coupling coefficient matrix, the distribution network data, the environmental prediction data, the distributed energy output data, and the user electricity consumption behavior data are respectively input into the deep reinforcement learning model to obtain the initial reconfiguration scheme of the distribution network output by the deep reinforcement learning model.

14. The cross-regional coordinated control device for power distribution networks according to claim 8, characterized in that, The local decision-making scheme based on multiple transformer substations and the initial reconfiguration scheme of the distribution network determine the target reconfiguration scheme of the distribution network, including: When the initial reconfiguration scheme of the distribution network conflicts with the local decision-making scheme of the first distribution area, the initial reconfiguration scheme of the distribution network is determined as the target reconfiguration scheme of the distribution network of the first distribution area. In the event of an emergency in the first distribution area, the first distribution area shall prioritize handling the emergency before executing the distribution network target reconfiguration scheme; the emergency event includes voltage over-limit events and / or equipment overload events, and the first distribution area is any one of the plurality of distribution areas.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the cross-regional coordinated control method for distribution networks as described in any one of claims 1 to 7.

16. A machine-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the cross-regional coordinated control method for distribution networks as described in any one of claims 1 to 7.

17. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the cross-regional coordinated control method for distribution networks as described in any one of claims 1 to 7.

Images